You are on page 1of 8

MULTIAXIAL ONE WAY SHAPE MEMORY EFFECT AND SUPERELASTICITY

Karine Taillarda, Sylvain Callochb, Shabnam Arbab Chiranic, Christian Lexcellentd.


a
LMT Cachan 61 avenue du Prsident Wilson 94235 Cachan cedex FRANCE
Karine.Taillard@lmt.ens-cachan.fr
b
LMSN Brest FRANCE
c
LI2 Brest FRANCE
d
LMARC Besanon FRANCE

ABSTRACT

The specific behavior of Shape Memory Alloys (SMA) is due to a solid-solid transformation called martensitic transformation
[1]. This reversible transformation consists mainly in a shear without volume change and is activated either by stress or
temperature. This partition between austenite (high temperature phase) and martensite (low temperature phase) is responsible
for the specific properties of SMA: the superelastic behavior (isothermal) and the one way shape memory effect (non-
isothermal).
The superelastic effect is obtained for fully austenitic SMA: loaded up to 5% strain, a sample recovers its initial shape after
unloading with a hysteretic loop. The one way shape memory effect is obtained when a martensitic SMA, which is apparently
plastically deformed, recovers its initial shape by simple heating.
Superelasticity and one way shape memory effect are useful for several three-dimensional applications. Despite all these
phenomena are well known and modeled for one-dimensional mechanical solicitations, the 3D behavior, and especially the
one way shape memory effect, remains quite unexplored (Tokuda et al. [2]). Actually the development of complex 3D
applications requires time consuming iterations and expensive prototypes. Predictive phenomenological models are
consequently crucial objectives for the design and dimensioning of such SMA structures.
Therefore, we perform several multi axial proportional and non-proportional, isothermal and non-isothermal tests. This
database will be used to build a phenomenological model within the framework of irreversible processes.

Introduction

Among the various properties of SMA, several specificities are of great importance for their description. The first one is the
coupling between mechanical and thermal behavior that is responsible for instance for the one way shape memory effect.
Otherwise, the peculiar behavior during non proportional 3D mechanical loadings is due to the specific mechanism of
martensite or R-phase reorientation. Moreover, the well-known tension-compression asymmetry has to be considered.
Therefore the description of the specific behavior of SMA requires non isothermal models, able to take into account 3D
proportional and non proportional loadings and the tension-compression asymmetry.

Many constitutive macro-models describing the behavior of shape memory alloys were developed in the last ten years [3, 4, 5].
But no one of them takes into account all the above inherent properties of SMA. Furthermore, very few of them are compared
with non isothermal, non proportional 3D tests results. The main reason of it is that very few complete test databases are
available for SMA.

The present paper deals with experimental investigations on a textured nickel-titanium SMA. Both isothermal and anisothermal
bi axial proportional and non proportional tests are performed. This database is then analyzed and an equivalent stress is
defined.

Material

The polycrystalline nickel-titanium (Ti-Ni 55.4 wt%) used for the tests was furnished by Nitinol Devices and Components. The
samples have been obtained by drawing, then machined and heat treated (2 minutes in salt bath at 480C ) by the Nitifrance
company. The transformation temperatures at free stress state have been determined using differential scanning calorimetry
(Figure 1). Nevertheless, the experimental device was not able to reach sufficiently low temperature, to determine all
transformation temperatures, but only R phase start temperature, Rs = 26C, R phase finish temperature , Rf = 11C and
austenite finish temperature, Af = 25C.

Figure 1. DSC analysis of the Ni-Ti

Experimental
The tests have been carried out on a MTS axial-torsion hydraulic testing machine. The axial and rotary actuators have a load
capacity of 100 kN and 1.2 kN.m, respectively. Axial and shear strains were measured using a rosette-type strain gauge glued
on the external diameter of the specimen. The samples are thin walled tubes; their geometry is given on figure 2.

Figure 2. Tension-torsion thin walled tube specimen

A thermal chamber is also used for temperature regulation.

Isothermal case: superelasticity


The main objective of this part is to obtain an experimental yield stress surface using so called detection tests. In this work,
yield surface is defined conventionally at a certain off-set of transformation strain (e.g., 0.1%). More details concerning the
experimental procedure can be found in [6].
Nine proportional loading paths have been carried out at 50C. The dots on Figure 3 shows the experime ntal shape of the
transformation yield surface in the (, ) plane. Notice the importance of the tension-compression assymmetry.
600
NiTi 50C

Shear stress (MPa) 400

200

-200

-400

-600
-800 -600 -400 -200 0 200 400 600
Axial stress (MPa)
Figure 3. Tension-torsion yield stress surface

To be able to describe this experimental yield stress surface an equivalent stress is defined. In this proposition, the isotropic
equivalent stress defined by Bouvet et al. [7] for isotropic SMA is generalized to take into account the anisotropy of the
material. A dilated stress tensor, denoted by ~ , is defined as a linear transformation of by:

~ = D. (1)

where D contains constant material parameters.

This affine transformation D is chosen such as the Hills hyper ellipsoid becomes a hyper sphere in the dilated stress space.

The form of D is as following:

2 A 2B 2 1 A B 1 A B 1
cos ( )+ sin ( )+ cos 2 ( ) sin 2 ( )+ cos 2 ( ) sin 2 ( )+
2
0 0 0
3 3 3 3 3 3 3 3 3
A B 2 A B 2 1 A B 2 B A 2 1
sym + cos ( )+ + sin ( )+ cos ( )+ sin ( )+ 0 0 0
6 2 2 6 3 6 2 6 2 3 (2)

A B A B 2 1
+ .cos ( )+ + .sin ( )+
2
1 sym sym 0 0 0
D= 6 2 2 6 3

0 0 0 L 0 0


0 0 0 0 M 0


0 0 0 0 0 N

with:
A 2 = . cos 2 ( ) + .sin 2 ( ) + . cos( ).sin( )
B 2 = .sin 2 ( ) + . cos 2 ( ) . cos( ).sin( )
1
= . arctan( ) (3)
2

=
1
2
[(D11 D12 )2 + (D11 D13 )2 + (D12 D13 )2 ]
and:

3
= .( F + H )
2
1 1
= F + H + 2.G (4)
2 2
= 3.( H F )
where F, G, H, L, M, N are the Hills criterion coefficients:

F( 11 22 )2 + G( 22 33 )2 + H( 33 11 )2 + 2L 12
2
+ 2 M 23
2
+ 2N 132 = 1 (5)

The angle denoted , Eq. (3.10), corresponds to the inclination of the Hills hyper ellipsoid in the stress space. Based on this
linear transformation, an equivalent stress for anisotropic material, eq ani1 , is simply defined by substituting ~ for in the
equivalent stress defined by Bouvet et al. [7] for isotropic SMA.
eqani1 = eq ani1 (~, y~ ) = ~.g ( y~ ) (6)

cos 1 (1 a.(1 y~ ))
g ( y~ ) = cos
(7)
3
where a is a material parameter permitting to fit the tension-compression asymmetry. An identification procedure [8] allows
5 1 5 1
determining the seven required parameters: F=G=H=0.26910 (MPa ), L=M=N=0.29710 (MPa ) and a =0.97. Then, the
so identified equivalent stress is compared to the experimental yield stress surface. The agreement between these two sets of
data is good. The same identification for other anisotropic SMA for various loadings plans [8] shows that this anisotropic yield
surface is flexible enough to describe the whole set of experimental data.

Anisothermal case: one way shape memory effect


The same detection tests were performed at -10C, w hen the material is mainly constituted of R-phase. In this case the
linearity lost denotes the R-phase reorientation. Thirteen proportional loading paths were performed in tension-compression-
torsion. The experimental results are given by the dots on figure 4. The tension-compression asymmetry is very high (yield
stress in compression is two times higher than the yield stress in tension) and this results shows that this properties is of
crucial importance in SMA models.
40
NiTi -10C

30

Shear stress (MPa) 20

10

-10

-20

-30
-50 -40 -30 -20 -10 0 10 20 30

Axial stress (MPa)

Figure 4. Tension-torsion yield stress surface

This experimental yield stress surface for reorientation is compared with the equivalent stress already identified for the
superelasticity. The same parameters are taken, and the equivalent stress is reported on figure 4 (continuous line). The
agreement between these two curves is very good. Consequently, the same equivalent stress can be used for 3D simulations
for all temperatures.
To complete the database of non isothermal test, several one way shape memory effects tests were performed. The process
of the tests is always the same: cooling from 50C to -10C, mechanical strain controlled proportional loading then mechanical
stress controlled unloading and finally heating from -10C to 50C. These tests highlight the one way shape memory effect: the
material recovers completely its initial shape without residual strain. Results in pure tension, pure torsion, pure compression
and combined tension-torsion compression-torsion loadings are presented in the following figures (5 to 8).

Figure 5. One way shape memory effect for pure tension and compression
Figure 6. One way shape memory effect for pure torsion

Figure 7. One way shape memory effect for combined tension-torsion loading
Figure 8. One way shape memory effect for combined compression-torsion loading

Conclusion

This kind of test database is the only one established for various temperatures and 2D loadings for the same SMA. This
database has now to be used for the construction, identification and validation of a macroscopic model for SMA behavior.
The importance of the tension-compression asymmetry for Ni-Ti SMA has been highlighted and a proposition of equivalent
stress has been validated to take it into account. This equivalent stress is consistent whatever the temperature is.

References

1. Patoor, E. and Berveiller, M. Technologie des Alliages Mmoire de Forme, Herms (1994).
2. Tokuda, M., Sittner, P., Takakura, M. and Ye, M., Experimental study on performances in Cu-based shape memory alloy
under multiaxial loading conditions, Mat. Sc. Res. Int. , 1, 260-265 (1995).
3. Helm, D. and Haupt, P., Shape memory behavior: modeling within continuum thermomechanics Int. J. of Sol. and
Struct., 40, 827-849 (2003).
4. Juhasz, L., Schnack, E., Hesebeck, O. and Andr, H., Macroscopic modeling of shape memory alloys under non-
proportional thermo-mechanical loadings, J. of Int. Mat. Syst. and Struct., 13, 825-836 (2002).
5. Lexcellent, C., Boubakar, M. L., Bouvet, C. et Calloch, S., About modeling the shape memory alloy behavior based on the
phase transformation surface identification under proportional loading and anisothermal conditions, Int. J. of Sol. and
Struct., 43, 613-626 (2006).
6. Bouvet, C., Calloch, S. and Lexcellent, C., A phenomenological model for pseudoelasticity of shape memory alloys under
multi-axial proportional and non-proportional loadings, Eur. J. Mech. A Solids, 23, 37-61 (2004).
7. Bouvet, C., Calloch, S. and Lexcellent, C., Mechanical behavior of a Cu-Al-Be shape memory alloy under multiaxial
proportional and non proportional loadings, In Transactions of the ASME, 112-124 (2002).
8. Taillard, K., Arbab-Chirani, S., Calloch, S. and Lexcellent, C., Relation between martensite volume fraction and
transformation strain for isotropic and anisotropic shape memory alloys, Mech. of Mat., submitted (2006).

You might also like